The aim of the proposed research is to investigate the feasibility, acceptability, and efficacy of using a small lightweight portable computer to reduce the cost of pychotherapy. This aim will be tested in the treatment of obesity which is a significant and prevalent health problem, and for which a well tested behavioral approach to treatment, easily adaptable to computer use, exists. Studies of computer-assisted psychotherapy to date, while promising, have been relatively small. Thus, a larger scale randomized study is now proposed, comparing the relative efficacy of therapist conducted and computer-assisted therapy. In addition, the level of group support needed for computer-assisted therapy, and the effect of varying the length of access to the computer will be examined. 342 women, between 20% and 40% overweight, will be randomly allocated to one of three groups: A therapist conducted weight reduction program (12 sessions over 20 weeks); Computer-assisted therapy with group support (5 sessions over 20 weeks); and unsupported computer-assisted therapy (1 introductory session only). Half of each computer-assisted group will be randomly assigned to receive the computer for 12 weeks and half for 20 weeks. Participants in the study will be followed for one year. The major outcome measure will be weight change. In addition, self-reported measures of eating behavior and activity, selected psychological measures, and a measure of treatment acceptability will be used for all participants. Finally, in a randomly selected subsample of participants, direct measures of eating behavior and activity levels will be obtained. These measures will allow comparison not only of weight loss between the groups, but also of the behavior changes induced by computer-assisted and therapist conducted treatment. The cost-effectiveness of the different treatment conditions will also be determined. The use of the computer as a data collection device will also be exploited in this study, using the frequently collected caloric consumption and activity data to develop a model of weight loss, that may allow smaller sample sizes to be used in future outcome studies in this area. Results from this study may also be generalizable to other problems for which computer-assisted psychotherapy may be feasible.